Yoshua Bengio, a pioneer in deep learning, has launched LawZero, aiming to promote safer AI systems. One initiative, Scientist AI, will function as a non-agentic AI intended to monitor and set guardrails for existing AI systems. By focusing on explaining observations rather than taking action, it seeks to enhance AI safety and mitigate overconfidence, which often leads to errors in AI behavior. Bengio hopes this approach can help avoid risks while harnessing the benefits of AI technology, urging stakeholders to reassess the future of AI development.
"Ultimately, focusing on non-agentic AI may enable the benefits of AI innovation while avoiding the risks associated with the current trajectory," the paper said.
"This system is designed to explain the world from observations, as opposed to taking actions in it to imitate or please humans," a paper on Scientist AIs said.
Collection
[
|
...
]